lumencode 1.3.2 → 1.3.3

package/lib/attribution.js

@@ -99,6 +99,8 @@ export function aggregateAttribution(items = []) {
     unknown: 0,
     human: 0,
     excluded: 0,
+    mergeCommits: 0,
+    mergeCommitLines: 0,
     confirmedAILines: 0,
     probableAILines: 0,
     possibleAILines: 0,
@@ -112,6 +114,12 @@ export function aggregateAttribution(items = []) {
 
   for (const item of items || []) {
     const classified = item?.classification ? item : classifyAttribution(item);
+    // 合并提交不计入统计，避免稀释 AI 占比
+    if (item?.reason === 'human_merge') {
+      summary.mergeCommits = (summary.mergeCommits || 0) + 1;
+      summary.mergeCommitLines = (summary.mergeCommitLines || 0) + (item?.added || item?.linesAdded || 0) + (item?.deleted || item?.linesDeleted || 0);
+      continue;
+    }
     const lines = (item?.added || item?.linesAdded || 0) + (item?.deleted || item?.linesDeleted || 0);
     summary.totalItems++;
     summary.totalLinesChanged += lines;

package/lib/git.js

@@ -550,7 +550,10 @@ export function computeAIContribution(commits, toolFilter = null, options = {})
   const attributionOptions = resolveAttributionOptions(options.attribution || options);
   const confidenceWeights = attributionOptions.confidenceWeights;
   const allCommits = commits || [];
+  const isMergeCommit = c => c.attributionType === 'human_merge';
   for (const c of allCommits) {
+    // 合并提交不计入总行数分母，避免稀释 AI 占比
+    if (isMergeCommit(c)) continue;
     totalLinesAdded += c.linesAdded || 0;
     totalLinesDeleted += c.linesDeleted || 0;
   }
@@ -807,14 +810,14 @@ function sanitizeArg(s) {
   return String(s || '').replace(/[`$"\\|;&<>!\n\r]/g, '');
 }
 
-function buildGitArgs(since, until) {
-  const sinceFull = since.includes('T') ? since : since + 'T00:00:00';
-  const safeSince = sanitizeArg(sinceFull);
-  const safeUntil = sanitizeArg(until);
-  // 格式：哨兵行(subject) → body 行(可多行) → ENDBODY 行 → numstat 行
-  const pretty = `--pretty=format:"${COMMIT_SENTINEL}%H|%ad|%ae|%s%n%B${BODY_END}"`;
-  return `--all --no-renames ${pretty} --date=iso-strict --numstat --since="${safeSince}" --until="${safeUntil}"`;
-}
+function buildGitArgs(since, until) {
+  const sinceFull = since.includes('T') ? since : since + 'T00:00:00';
+  const safeSince = sanitizeArg(sinceFull);
+  const safeUntil = sanitizeArg(until);
+  // 格式：哨兵行(subject) → body 行(可多行) → ENDBODY 行 → numstat 行
+  const pretty = `--pretty=format:"${COMMIT_SENTINEL}%H|%ad|%ae|%s%n%B${BODY_END}"`;
+  return `--all --no-renames ${pretty} --date=iso-strict --numstat --since="${safeSince}" --until="${safeUntil}"`;
+}
 
 function mergeGitStats(target, source) {
   target.commits += source.commits;
@@ -837,66 +840,66 @@ function mergeGitStats(target, source) {
   // filesChanged 在 merge 完后由 finalize 重新计算（跨 repo 去重）
 }
 
-function recomputeFilesChanged(stats) {
-  const set = new Set();
-  for (const c of stats.commitList || []) {
-    for (const f of c.files || []) set.add((c.repo || '') + '::' + f.path);
-  }
-  stats.filesChanged = set.size;
-}
-
-function recomputeStatsFromCommitList(stats) {
-  stats.commits = 0;
-  stats.filesChanged = 0;
-  stats.linesAdded = 0;
-  stats.linesDeleted = 0;
-  stats.commitsByDate = {};
-  stats.linesByDate = {};
-
-  for (const c of stats.commitList || []) {
-    const dateKey = c.dateLocal || (c.date || '').slice(0, 10);
-    stats.commits++;
-    stats.commitsByDate[dateKey] = (stats.commitsByDate[dateKey] || 0) + 1;
-    if (!stats.linesByDate[dateKey]) stats.linesByDate[dateKey] = { added: 0, deleted: 0, files: 0 };
-    stats.linesByDate[dateKey].added += c.linesAdded || 0;
-    stats.linesByDate[dateKey].deleted += c.linesDeleted || 0;
-    stats.linesByDate[dateKey].files += (c.files || []).length;
-    stats.linesAdded += c.linesAdded || 0;
-    stats.linesDeleted += c.linesDeleted || 0;
-  }
-  recomputeFilesChanged(stats);
-}
-
-function markAuthorOwnership(stats, expectedAuthor) {
-  const normalizedExpected = (expectedAuthor || '').toLowerCase();
-  for (const c of stats.commitList || []) {
-    c.expectedAuthor = expectedAuthor || null;
-    c.authorMatchesConfig = normalizedExpected
-      ? (c.author || '').toLowerCase() === normalizedExpected
-      : null;
-  }
-}
-
-function hasLocalSessionEvidence(commit) {
-  if (!commit.sessionId) return false;
-  if (commit.sessionAttribution === 'strong') return true;
-  return (commit.aiEvidenceDetails?.matchedFileCount || 0) > 0;
-}
-
-function filterCommitsForUser(stats) {
-  const commits = stats.commitList || [];
-  const hasAuthorOwnershipMetadata = commits.some(c => c.expectedAuthor || c.authorMatchesConfig !== undefined);
-
-  for (const c of commits) {
-    c.countedForUser = !hasAuthorOwnershipMetadata
-      || c.authorMatchesConfig === true
-      || hasLocalSessionEvidence(c);
-  }
-
-  if (!hasAuthorOwnershipMetadata) return;
-  stats.commitList = commits.filter(c => c.countedForUser);
-  recomputeStatsFromCommitList(stats);
-}
+function recomputeFilesChanged(stats) {
+  const set = new Set();
+  for (const c of stats.commitList || []) {
+    for (const f of c.files || []) set.add((c.repo || '') + '::' + f.path);
+  }
+  stats.filesChanged = set.size;
+}
+
+function recomputeStatsFromCommitList(stats) {
+  stats.commits = 0;
+  stats.filesChanged = 0;
+  stats.linesAdded = 0;
+  stats.linesDeleted = 0;
+  stats.commitsByDate = {};
+  stats.linesByDate = {};
+
+  for (const c of stats.commitList || []) {
+    const dateKey = c.dateLocal || (c.date || '').slice(0, 10);
+    stats.commits++;
+    stats.commitsByDate[dateKey] = (stats.commitsByDate[dateKey] || 0) + 1;
+    if (!stats.linesByDate[dateKey]) stats.linesByDate[dateKey] = { added: 0, deleted: 0, files: 0 };
+    stats.linesByDate[dateKey].added += c.linesAdded || 0;
+    stats.linesByDate[dateKey].deleted += c.linesDeleted || 0;
+    stats.linesByDate[dateKey].files += (c.files || []).length;
+    stats.linesAdded += c.linesAdded || 0;
+    stats.linesDeleted += c.linesDeleted || 0;
+  }
+  recomputeFilesChanged(stats);
+}
+
+function markAuthorOwnership(stats, expectedAuthor) {
+  const normalizedExpected = (expectedAuthor || '').toLowerCase();
+  for (const c of stats.commitList || []) {
+    c.expectedAuthor = expectedAuthor || null;
+    c.authorMatchesConfig = normalizedExpected
+      ? (c.author || '').toLowerCase() === normalizedExpected
+      : null;
+  }
+}
+
+function hasLocalSessionEvidence(commit) {
+  if (!commit.sessionId) return false;
+  if (commit.sessionAttribution === 'strong') return true;
+  return (commit.aiEvidenceDetails?.matchedFileCount || 0) > 0;
+}
+
+function filterCommitsForUser(stats) {
+  const commits = stats.commitList || [];
+  const hasAuthorOwnershipMetadata = commits.some(c => c.expectedAuthor || c.authorMatchesConfig !== undefined);
+
+  for (const c of commits) {
+    c.countedForUser = !hasAuthorOwnershipMetadata
+      || c.authorMatchesConfig === true
+      || hasLocalSessionEvidence(c);
+  }
+
+  if (!hasAuthorOwnershipMetadata) return;
+  stats.commitList = commits.filter(c => c.countedForUser);
+  recomputeStatsFromCommitList(stats);
+}
 
 // ── async versions (server) with cache ──
 
@@ -916,26 +919,26 @@ function evictGitCache() {
   }
 }
 
-async function getGitStatsAsync(repoPath, since, until, author = null) {
-  const cacheKey = `${repoPath}|${since}|${until}|${CACHE_VERSION}`;
-  const cached = gitCache.get(cacheKey);
-  if (cached && Date.now() - cached.ts < GIT_CACHE_TTL) return cached.stats;
+async function getGitStatsAsync(repoPath, since, until, author = null) {
+  const cacheKey = `${repoPath}|${since}|${until}|${CACHE_VERSION}`;
+  const cached = gitCache.get(cacheKey);
+  if (cached && Date.now() - cached.ts < GIT_CACHE_TTL) return cached.stats;
 
   try {
     await execAsync('git rev-parse --git-dir', { cwd: repoPath });
   } catch {
     return emptyResult();
   }
-
-  try {
-    const output = await execAsync(`git log ${buildGitArgs(since, until)}`, {
-      cwd: repoPath, encoding: 'utf-8', maxBuffer: 50 * 1024 * 1024,
-    });
-    const stats = parseGitLogOutput(output, repoPath);
-    markAuthorOwnership(stats, author);
-    gitCache.set(cacheKey, { stats, ts: Date.now() });
-    evictGitCache();
-    return stats;
+
+  try {
+    const output = await execAsync(`git log ${buildGitArgs(since, until)}`, {
+      cwd: repoPath, encoding: 'utf-8', maxBuffer: 50 * 1024 * 1024,
+    });
+    const stats = parseGitLogOutput(output, repoPath);
+    markAuthorOwnership(stats, author);
+    gitCache.set(cacheKey, { stats, ts: Date.now() });
+    evictGitCache();
+    return stats;
   } catch {
     return emptyResult();
   }
@@ -1131,38 +1134,38 @@ function sortAttributionCandidates(candidates) {
   });
 }
 
-function candidateFromSession(commit, session, distanceMs) {
-  const overlap = computeFileOverlap(session.touchedFiles || [], commit.files || []);
-  return scoreSessionCandidate(commit, session, {
-    distanceMs,
+function candidateFromSession(commit, session, distanceMs) {
+  const overlap = computeFileOverlap(session.touchedFiles || [], commit.files || []);
+  return scoreSessionCandidate(commit, session, {
+    distanceMs,
     fileOverlapRatio: overlap.fileOverlapRatio,
     matchedFiles: overlap.matchedFiles,
     projectMatches: true,
-  });
-}
-
-function getStepSessionIdCandidates(sessionId, session) {
-  if (!sessionId) return [];
-  const candidates = [sessionId];
-  if (sessionId.includes(':')) return candidates;
-
-  const originByTool = {
-    claude: 'claude_code',
-    codex: 'codex_cli',
-  };
-  const origin = originByTool[session?.primaryTool];
-  if (origin) {
-    candidates.push(`${origin}:${sessionId}`);
-  } else {
-    candidates.push(`claude_code:${sessionId}`, `codex_cli:${sessionId}`);
-  }
-
-  return [...new Set(candidates)];
-}
-
-const BASH_GIT_COMMIT_RE = /\bgit\s+commit\b/i;
-const STRONG_WINDOW_BEFORE_MS = 30 * 1000;       // 30s before bash invocation
-const STRONG_WINDOW_AFTER_MS = 5 * 60 * 1000;    // 5min after
+  });
+}
+
+function getStepSessionIdCandidates(sessionId, session) {
+  if (!sessionId) return [];
+  const candidates = [sessionId];
+  if (sessionId.includes(':')) return candidates;
+
+  const originByTool = {
+    claude: 'claude_code',
+    codex: 'codex_cli',
+  };
+  const origin = originByTool[session?.primaryTool];
+  if (origin) {
+    candidates.push(`${origin}:${sessionId}`);
+  } else {
+    candidates.push(`claude_code:${sessionId}`, `codex_cli:${sessionId}`);
+  }
+
+  return [...new Set(candidates)];
+}
+
+const BASH_GIT_COMMIT_RE = /\bgit\s+commit\b/i;
+const STRONG_WINDOW_BEFORE_MS = 30 * 1000;       // 30s before bash invocation
+const STRONG_WINDOW_AFTER_MS = 5 * 60 * 1000;    // 5min after
 
 function toMs(iso) {
   if (!iso) return NaN;
@@ -1346,11 +1349,11 @@ export function attachCommitsToSessions(sessions, commitList) {
 }
 
 // 一次性收尾：跑 attribution + 三个聚合
-export async function finalizeGitStats(merged, sessions = [], options = {}) {
-  if (!merged) return merged;
-  const attributionOptions = resolveAttributionOptions(options.attribution || options);
-  const stepTrackingOptions = options.stepTracking || {};
-  const fileOverrides = loadAttributionOverrides();
+export async function finalizeGitStats(merged, sessions = [], options = {}) {
+  if (!merged) return merged;
+  const attributionOptions = resolveAttributionOptions(options.attribution || options);
+  const stepTrackingOptions = options.stepTracking || {};
+  const fileOverrides = loadAttributionOverrides();
   const inputOverrides = options.overrides || {};
   const mergedOverrides = {
     commits: { ...fileOverrides.commits, ...(inputOverrides.commits || {}) },
@@ -1378,69 +1381,69 @@ export async function finalizeGitStats(merged, sessions = [], options = {}) {
     }
   }
 
-  // Step 1.5: Enrich commits with line-level step blame when available
-  const stepTrackers = new Map();
-  if (stepTrackingOptions.enabled !== false) try {
-    const { StepTracker } = await import('./step-tracker.js');
-    const projectRoots = [...new Set((sessions || []).map(s => s.project).filter(Boolean))];
-    // Also check repo paths from commits
-    for (const c of merged.commitList || []) {
-      if (c.repo) projectRoots.push(c.repo);
-    }
-    for (const root of [...new Set(projectRoots.map(normalizePathForGit))]) {
-      if (!root) continue;
-      const tracker = new StepTracker(root, {
-        dbPath: stepTrackingOptions.dbPath,
-        maxFileSize: stepTrackingOptions.maxFileSize,
-      });
-      if (await tracker.isAvailableAsync()) {
-        await tracker.open();
-        stepTrackers.set(root, tracker);
-      }
-    }
-  } catch {
-    for (const tracker of stepTrackers.values()) tracker.close();
-    stepTrackers.clear();
-  }
-
-  if (stepTrackers.size > 0) {
-    for (const c of merged.commitList || []) {
-      if (!c.sessionId) continue;
-      const candidateRoots = [
-        c.repo,
-        sessionsById.get(c.sessionId)?.project,
-      ].filter(Boolean);
-      let stepTracker = null;
-      for (const candidateRoot of candidateRoots) {
-        const normalizedCandidate = normalizePathForGit(candidateRoot);
-        for (const [root, tracker] of stepTrackers.entries()) {
-          if (projectMatchesFromGitPaths(root, normalizedCandidate)) {
-            stepTracker = tracker;
-            break;
-          }
-        }
-        if (stepTracker) break;
-      }
-      if (!stepTracker && stepTrackers.size === 1) {
-        stepTracker = stepTrackers.values().next().value;
-      }
-      if (!stepTracker) continue;
-      try {
-        const session = sessionsById.get(c.sessionId);
-        for (const stepSessionId of getStepSessionIdCandidates(c.sessionId, session)) {
-          const lineBlame = stepTracker.getLineAttributionForCommit({
-            ...c,
-            sessionId: stepSessionId,
-          });
-          if (lineBlame) {
-            c.lineBlame = lineBlame;
-            break;
-          }
-        }
-      } catch { /* best effort */ }
-    }
-    for (const tracker of stepTrackers.values()) tracker.close();
-  }
+  // Step 1.5: Enrich commits with line-level step blame when available
+  const stepTrackers = new Map();
+  if (stepTrackingOptions.enabled !== false) try {
+    const { StepTracker } = await import('./step-tracker.js');
+    const projectRoots = [...new Set((sessions || []).map(s => s.project).filter(Boolean))];
+    // Also check repo paths from commits
+    for (const c of merged.commitList || []) {
+      if (c.repo) projectRoots.push(c.repo);
+    }
+    for (const root of [...new Set(projectRoots.map(normalizePathForGit))]) {
+      if (!root) continue;
+      const tracker = new StepTracker(root, {
+        dbPath: stepTrackingOptions.dbPath,
+        maxFileSize: stepTrackingOptions.maxFileSize,
+      });
+      if (await tracker.isAvailableAsync()) {
+        await tracker.open();
+        stepTrackers.set(root, tracker);
+      }
+    }
+  } catch {
+    for (const tracker of stepTrackers.values()) tracker.close();
+    stepTrackers.clear();
+  }
+
+  if (stepTrackers.size > 0) {
+    for (const c of merged.commitList || []) {
+      if (!c.sessionId) continue;
+      const candidateRoots = [
+        c.repo,
+        sessionsById.get(c.sessionId)?.project,
+      ].filter(Boolean);
+      let stepTracker = null;
+      for (const candidateRoot of candidateRoots) {
+        const normalizedCandidate = normalizePathForGit(candidateRoot);
+        for (const [root, tracker] of stepTrackers.entries()) {
+          if (projectMatchesFromGitPaths(root, normalizedCandidate)) {
+            stepTracker = tracker;
+            break;
+          }
+        }
+        if (stepTracker) break;
+      }
+      if (!stepTracker && stepTrackers.size === 1) {
+        stepTracker = stepTrackers.values().next().value;
+      }
+      if (!stepTracker) continue;
+      try {
+        const session = sessionsById.get(c.sessionId);
+        for (const stepSessionId of getStepSessionIdCandidates(c.sessionId, session)) {
+          const lineBlame = stepTracker.getLineAttributionForCommit({
+            ...c,
+            sessionId: stepSessionId,
+          });
+          if (lineBlame) {
+            c.lineBlame = lineBlame;
+            break;
+          }
+        }
+      } catch { /* best effort */ }
+    }
+    for (const tracker of stepTrackers.values()) tracker.close();
+  }
 
   // Step 1.6: compute developer behavioral baselines
   const authorBaselines = computeAuthorBaseline(merged.commitList);
@@ -1548,21 +1551,21 @@ export async function finalizeGitStats(merged, sessions = [], options = {}) {
   }
 
   // Step 2.5: composite continuous scoring for all commits
-  for (const c of merged.commitList || []) {
-    c.aiScore = computeContinuousScore(c, attributionOptions);
-    const mappedConfidence = scoreToConfidence(c.aiScore, attributionOptions);
-    // Only override if no explicit signature and continuous score disagrees
-    if (c.attributionType !== 'explicit') {
+  for (const c of merged.commitList || []) {
+    c.aiScore = computeContinuousScore(c, attributionOptions);
+    const mappedConfidence = scoreToConfidence(c.aiScore, attributionOptions);
+    // Only override if no explicit signature and continuous score disagrees
+    if (c.attributionType !== 'explicit') {
       c.aiConfidence = pickHigherConfidence(c.aiConfidence, mappedConfidence);
       c.isAI = isCountedAIConfidence(c.aiConfidence);
-      c.aiAssisted = c.aiConfidence !== AI_CONFIDENCE.NONE;
-    }
-  }
-
-  filterCommitsForUser(merged);
-
-  const attributionItems = [];
-  for (const c of merged.commitList || []) {
+      c.aiAssisted = c.aiConfidence !== AI_CONFIDENCE.NONE;
+    }
+  }
+
+  filterCommitsForUser(merged);
+
+  const attributionItems = [];
+  for (const c of merged.commitList || []) {
     const commitOverride = mergedOverrides.commits[c.hash] || null;
     const fileOverride = (c.files || []).find(f => mergedOverrides.files[`${c.hash}:${f.path}`]);
     const fileOverrideValue = fileOverride ? mergedOverrides.files[`${c.hash}:${fileOverride.path}`] : null;

package/lib/report.js

@@ -167,7 +167,11 @@ function buildAttributionSummaryLine(summary) {
   const confirmedPct = Math.round((summary.confirmedAILines / total) * 100);
   const upperPct = Math.round(((summary.confirmedAILines + summary.probableAILines + summary.possibleAILines) / total) * 100);
   const unknownPct = Math.round((summary.unknownLines / total) * 100);
-  return `AI 归因汇总：确认 AI ${confirmedPct}% / 可能 AI 上限 ${upperPct}% / 未知 ${unknownPct}%。`;
+  let line = `AI 归因汇总：确认 AI ${confirmedPct}% / 可能 AI 上限 ${upperPct}% / 未知 ${unknownPct}%。`;
+  if (summary.mergeCommits > 0) {
+    line += `（已排除 ${summary.mergeCommits} 个合并提交，共 ${formatInt(summary.mergeCommitLines)} 行）`;
+  }
+  return line;
 }
 
 function buildUnknownReasonLine(summary) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "lumencode",
-  "version": "1.3.2",
+  "version": "1.3.3",
   "description": "LumenCode — AI 编码助手使用报告工具，从 JSONL 日志和 Git 仓库提取 AI 贡献度、效率与使用指标，支持 Web 可视化和命令行两种模式",
   "type": "module",
   "bin": {